Our understanding of the properties of individual neurons and their role in brain computations has advanced significantly during the last few decades. However, we are still very far from understanding how large assemblies of cells interact to process information. Electrophysiology is the gold standard with unmatched temporal resolution, but is currently limited in terms of its ability record from every single neuron withina volume with cell-type specificity. Optical imaging provides a powerful alternative method, which enables localization of neurons in anatomical space and cell-type specificity via genetically encoded fluorescent markers. The current state-of-the-art in functional brain imaging is two-photon fluorescence laser-scanning microscopy. But this approach works best only on the surface of the brain, or transparent tissues and is not easily scalable. More generally, light scattering and absorption in tissue impose significant fundamental limits: in mammalian brains, accessible depths in vivo are restricted to superficial cortical regions, d1mm. Endoscopic methods developed to circumvent such restrictions impart significant damage to tissue above the imaging site given the large probe diameter (0.3 to >1 mm) and thus are quite limited (e.g. cannot be used to study cortical columns). Here we propose a novel paradigm for functional optical imaging that surmounts these limitations. It permits function- al imaging with cellular resolution in highly scattering brain tissue, enables complete coverage of all neurons within a target volume, and has long-term prospects for human applications. Our approach, which we term integrated neurophotonics, is based on distributing a dense 3-D lattice of emitter and detector pixels within the brain itself. These pixel arrays are embedded onto neurophotonic probes, realized as implantable, ultra narrow shanks that leverage recent advances in both integrated nanophotonics. Used with functional optical reporters (e.g. GCaMP6), one 25-shank probe module will be capable of recording the activity of all neurons within a 1- mm3 volume (~100,000 neurons) with single cell resolution. The methodology is scalable; multiple modules can be tiled to densely cover extended regions deep within the brain. It will ultimately permit simultaneous recording from millions of neurons at arbitrary positions and depths in the brain, to unveil the dynamics of complete neural networks - with single-cell resolution and cell-type specificity. Ultra-narrow neurophotonic probes will perturb brain tissue minimally, imposing negligible tissue displacement and minute local power dissipation. Importantly, they are readily producible though existing wafer-scale foundry (factory) based methods and thus will be widely available for use by the community. They will transform studies of circuit- level mechanisms of brain computation and neuropsychiatric disorders, and will accelerate drug discovery via high throughput in vivo screening. Our multi-disciplinary team spans all requisite expertise: nanotechnology and large-scale-integration for development of neurophotonic probe arrays (Roukes, Shepard), and in vivo testing and computational analysis (Tolias, Siapas).